Description: Pre-processing of ASV count table, including joining sequence count per sample information with taxonomic assignments, and environmental metadata. With sequenced blanks, use decontam to remove putative contaminant ASVs, then filter all samples by total sequence threshold. Visualize total sequences and ASVs before and after these steps. Additionally, assign ASV classifications based on distribution (i.e., cosmopolitan vs. resident, GR and Axial vs. MCR?). Final step saves QCed and classified ASV table as an R object in “input data” for main data analysis to be complete. See: https://shu251.github.io/microeuk-amplicon-survey/

Import exisiting sequence data

Datasets included: - Gorda Ridge 2019 cruise - Axial Seamount time series - 2013, 2014, & 2015 - Mid-Cayman Rise 2020 cruise

All data generated from extracted RNA, reverse transcribed to cDNA and amplified with primers that target the V4 hypervariable region on the 18S rRNA gene.

Analysis done with QIIME2, kept 40-60% of the sequences through the QC process and generated Amplicon Sequence Variants (ASVs) with DADA2. Taxonomic assignment done with vsearch using the PR2 database (v4.14) at 80% identity. See the seq-analysis directory for QIIME2 code.

Merged data from QIIME2 analysis

After determining ASVs for each sequence run, ASV tables were merged.

merged_tax <- read_delim("../data-input/taxonomy.tsv", delim = "\t")
merged_asv <- read_delim("../data-input/microeuk-merged-asv-table.tsv", delim = "\t", skip = 1)
# head(merged_tax)

Import metadata & reformat

Metadata (last update 26-01-2022), compiled from several different sources.

metadata <- read.delim("../data-input/samplelist-metadata.txt", na.strings = "")

Note that some of it is not numeric and will require downstream culling, as many ‘no data available’ samples exist.

metadata_formatted <- metadata %>% 
  mutate_all(as.character) %>%
  filter(Sample_or_Control == "Sample") %>% 
  filter(!(SAMPLETYPE == "Incubation")) %>% 
  filter(!(SAMPLETYPE == "Microcolonizer")) %>% 
  select(SAMPLE, VENT, COORDINATES, SITE, SAMPLEID, DEPTH, SAMPLETYPE, YEAR, TEMP = starts_with("TEMP"), pH, PercSeawater = starts_with("Perc"), Mg = starts_with("Mg"), H2 = starts_with("H2."), H2S = starts_with("H2S"), CH4 = starts_with("CH4"), ProkConc, Sample_or_Control)

Units for the variables are as follows: - Temp = Celsius - Percent Seawater = % - Mg = mmol/kg (or mM) - H2 = µmol/L (or µM) - H2S = mmol/L (or mM) - CH4 = µmol/L (or µM) - ProkConc = cells/ml

Remove samples from Gorda Ridge microcolonizers and from the FLP experiments (Gorda Ridge and Mid-Cayman Rise).

asv_wtax <- merged_asv %>%
  select(FeatureID = '#OTU ID', everything()) %>%
  pivot_longer(cols = !FeatureID,
               names_to = "SAMPLE", values_to = "value") %>%
  left_join(merged_tax, by = c("FeatureID" = "Feature ID")) %>%
  left_join(metadata_formatted) %>%
  filter(!grepl("Siders_", SAMPLE)) %>% 
  filter(SAMPLETYPE != "Incubation") %>% 
  filter(SAMPLETYPE != "Microcolonizer") %>%
  mutate(DATASET = case_when(
    grepl("_GR_", SAMPLE) ~ "GR",
    grepl("Gorda", SAMPLE) ~ "GR",
    grepl("_MCR_", SAMPLE) ~ "MCR",
    grepl("Axial", SAMPLE) ~ "Axial",
  TRUE ~ "Control or blank")) %>%
    separate(Taxon, c("Domain", "Supergroup",
                  "Phylum", "Class", "Order",
                  "Family", "Genus", "Species"), sep = ";", remove = FALSE) %>% 
  unite(SAMPLENAME, SITE, SAMPLETYPE, YEAR, VENT, SAMPLEID, sep = " ", remove = FALSE)

## Warning: Expected 8 pieces. Additional pieces discarded in 427968 rows [97, 98,
## 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114,
## 115, 116, ...].

## Warning: Expected 8 pieces. Missing pieces filled with `NA` in 432864 rows [1,
## 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, ...].

# View(asv_wtax)
# head(asv_wtax) ## Complete ASV table with full taxonomy names and annotated sample information

Total number of sequences

Barplots to show total number of sequences and total number of ASVs.

Total number of sequences and ASVs parallel each other. The Axial and Gorda Ridge data were run on the same sequence run, with Mid-Cayman Rise run on a separate MiSeq run - so the average number of sequences (and ASVs) varies between these two runs. A few samples have too few sequences, they will be removed below.

This newest version of PR2 has bacteria and archaea in it. Very, very few were assigned to this. Majority assigned to eukaryotes.

# head(asv_wtax)
library(viridis)
plot_grid(
  # Total number of ASVs
  asv_wtax %>% 
  filter(value > 0) %>% 
  filter(Sample_or_Control == "Sample") %>% 
  ggplot(aes(x = SAMPLENAME)) +
  geom_bar(stat = "count", width = 0.9) +
  labs(y = "Total ASVs per sample", x = "") +
  coord_flip() +
    scale_y_continuous(position = "right") +
  theme_linedraw() +
  facet_grid(DATASET ~ ., scale = "free", space = "free") +
  theme(axis.text.x = element_text(angle = 0, hjust = 1, vjust = 1),
        axis.text.y = element_text(angle = 0, hjust = 1, vjust = 1),
        strip.background = element_blank(), strip.text = element_text(color = "black")),
  asv_wtax %>% 
    filter(Sample_or_Control == "Sample") %>%
  group_by(SAMPLENAME, SITE, Domain, DATASET) %>% 
  summarise(SUM_SEQ_DOMAIN = sum(value)) %>% 
  ggplot(aes(x = SAMPLENAME, y = SUM_SEQ_DOMAIN, fill = Domain)) +
  geom_bar(stat = "identity", color = "black", width = 0.9) +
  labs(y = "Total sequences per sample", x = "") +
  coord_flip() +
  viridis::scale_fill_viridis(discrete = TRUE) +
  scale_y_continuous(position = "right") +
  theme_linedraw() +
  facet_grid(DATASET ~ ., scale = "free", space = "free") +
  theme(axis.text.x = element_text(angle = 0, hjust = 1, vjust = 1),
        axis.text.y = element_text(angle = 0, hjust = 1, vjust = 1),
        strip.background = element_blank(), strip.text = element_text(color = "black"),
        legend.position = "right"),
  ncol = 2, align = c("hv"), axis = c("lr"))

Table reporting total ASVs and sequences

table_raw_stats <- asv_wtax %>% filter(value > 0) %>%
       group_by(SAMPLENAME, DATASET, SITE) %>%
       summarise(SEQ_SUM = sum(value),
                 ASV_COUNT = n()) %>% 
  ungroup() %>% 
  gt(
    groupname_col = c("DATASET", "SITE"),
    rowname_col = "SAMPLENAME"
  )
table_raw_stats

	SEQ_SUM	ASV_COUNT
Axial - Axial
Axial Background 2015 Deep seawater BSW1500m	48657	494
Axial Plume 2015 Anemone Plume AnemonePlume	33599	474
Axial Vent 2013 Anemone FS891	38672	449
Axial Vent 2013 Boca FS905	190675	2174
Axial Vent 2013 Dependable FS900	52	4
Axial Vent 2013 El Guapo FS896	57160	804
Axial Vent 2013 Marker113 FS903	77473	685
Axial Vent 2013 Marker33 FS904	51169	473
Axial Vent 2013 N3Area FS898	53733	610
Axial Vent 2013 Skadi FS902	43916	40
Axial Vent 2014 Escargot FS910	73959	990
Axial Vent 2014 Marker113 FS906	33394	333
Axial Vent 2014 Marker33 FS908	59502	771
Axial Vent 2015 Marker113 FS915	63634	656
GR - GordaRidge
GordaRidge Background 2019 Deep seawater BSW020	5	1
GordaRidge Background 2019 Deep seawater BSW056	25095	498
GordaRidge Background 2019 Near vent BW Plume001	104106	1206
GordaRidge Background 2019 Shallow seawater BSW081	38196	572
GordaRidge Plume 2019 Candelabra Plume Plume036	57571	473
GordaRidge Plume 2019 Mt Edwards Plume Plume096	38604	587
GordaRidge Vent 2019 Candelabra Vent086	57369	721
GordaRidge Vent 2019 Candelabra Vent087	45802	663
GordaRidge Vent 2019 Candelabra Vent088	40719	631
GordaRidge Vent 2019 Mt Edwards Vent009	42591	245
GordaRidge Vent 2019 Mt Edwards Vent010	57564	558
GordaRidge Vent 2019 Mt Edwards Vent011	71430	665
GordaRidge Vent 2019 Sir Ventsalot Vent105	57998	456
GordaRidge Vent 2019 Sir Ventsalot Vent106	41250	654
GordaRidge Vent 2019 Sir Ventsalot Vent107	43230	590
GordaRidge Vent 2019 Venti Latte Vent039	45588	486
GordaRidge Vent 2019 Venti Latte Vent040	57186	688
GordaRidge Vent 2019 Venti Latte Vent041	64680	835
MCR - Piccard
Piccard Background 2020 BSW NA	139563	646
Piccard Plume 2020 Plume NA	123391	333
Piccard Vent 2020 LotsOShrimp NA	204065	687
Piccard Vent 2020 Shrimpocalypse NA	146104	706
MCR - VonDamm
VonDamm Background 2020 BSW NA	119010	864
VonDamm Plume 2020 Plume NA	151501	1165
VonDamm Vent 2020 ArrowLoop NA	115756	932
VonDamm Vent 2020 MustardStand NA	160115	56
VonDamm Vent 2020 OldManTree NA	144925	600
VonDamm Vent 2020 Rav2 NA	286292	1792
VonDamm Vent 2020 ShrimpHole NA	164954	961
VonDamm Vent 2020 WhiteCastle NA	177663	34
VonDamm Vent 2020 X18 NA	169301	638

gtsave(table_raw_stats, filename = "seq_asv_count_nonQC.html", path = "../output-tables/")

After removing contaminate ASVs below, I will set threshold of 10,000 sequences- if a sample has fewer than this, chuck it.

Decontaminate sequence library

Import sample description text file, import as phyloseq library, and remove potential contaminate ASVs and sequences. Catalog total number of ASVs and sequences removed from analysis.

# library(decontam); library(phyloseq)

Import as phyloseq objects

tax_matrix <- merged_tax %>% 
  select(FeatureID = `Feature ID`, Taxon) %>% 
  separate(Taxon, c("Domain", "Supergroup", 
                  "Phylum", "Class", "Order",
                  "Family", "Genus", "Species"), sep = ";", remove = FALSE) %>% 
  column_to_rownames(var = "FeatureID") %>% 
  as.matrix

## Warning: Expected 8 pieces. Additional pieces discarded in 8916 rows [1, 2, 5,
## 8, 9, 10, 11, 12, 16, 19, 21, 22, 23, 24, 26, 27, 28, 29, 30, 33, ...].

## Warning: Expected 8 pieces. Missing pieces filled with `NA` in 9018 rows [3, 4,
## 6, 7, 13, 14, 15, 17, 18, 20, 25, 31, 32, 38, 39, 42, 43, 44, 46, 47, ...].

asv_matrix <- merged_asv %>% 
  select(FeatureID = '#OTU ID', everything()) %>% 
  column_to_rownames(var = "FeatureID") %>% 
  as.matrix

# Align row names for each matrix
rownames(tax_matrix) <- row.names(asv_matrix)

# Set rownames of metadata table to SAMPLE information
row.names(metadata) <- metadata$SAMPLE

# Import asv and tax matrices
ASV = otu_table(asv_matrix, taxa_are_rows = TRUE)
TAX = tax_table(tax_matrix)
phylo_obj <- phyloseq(ASV, TAX)

# Import metadata as sample data in phyloseq
samplenames <- sample_data(metadata)

# join as phyloseq object
physeq_wnames = merge_phyloseq(phylo_obj, samplenames)
# colnames(ASV)

## Check
# physeq_wnames

Identify contaminant ASVs

# When "Control" appears in "Sample_or_Control column, this is a negative control"
sample_data(physeq_wnames)$is.neg <- sample_data(physeq_wnames)$Sample_or_Control == "Control"

# ID contaminants using Prevalence information
contam_prev <- isContaminant(physeq_wnames, 
                               method="prevalence", 
                               neg="is.neg", 
                               threshold = 0.5, normalize = TRUE)

## Warning in .is_contaminant(seqtab, conc = conc, neg = neg, method = method, :
## Removed 1 samples with zero total counts (or frequency).

## Warning in .is_contaminant(seqtab, conc = conc, neg = neg, method = method, :
## Removed 1 samples with zero total counts (or frequency).

# Report number of ASVs IDed as contaminants
table(contam_prev$contaminant)

## 
## FALSE  TRUE 
## 17878    56

0.5 - this threshold will ID contaminants in all samples that are more prevalent in negative controls than in positive samples.

As of Dec 30 2021: 56 ASVs deemed to be contaminant and will be removed.

Remove problematic ASVs

# Subset contaminant ASVs
contams <- filter(contam_prev, contaminant == "TRUE")
list_of_contam_asvs <- as.character(row.names(contams))
# length(list_of_contam_asvs)

taxa_contam <- as.data.frame(tax_matrix) %>% 
  rownames_to_column(var = "FeatureID") %>% 
  filter(FeatureID %in% list_of_contam_asvs)
# head(taxa_contam)

# View(asv_wtax)
asv_wtax_decon <- asv_wtax %>% 
  filter(!(FeatureID %in% list_of_contam_asvs)) %>% 
  filter(!(Sample_or_Control == "Control"))

tmp_orig <- (asv_wtax %>% filter(!(Sample_or_Control == "Control")))

# Stats on lost
x <- length(unique(tmp_orig$FeatureID)); x

## [1] 17934

y <- length(unique(asv_wtax_decon$FeatureID)); y

## [1] 17878

y-x

## [1] -56

100*((y-x)/x) # 56 total ASVs lost

## [1] -0.312256

a <- sum(tmp_orig$value);a #3.817 million

## [1] 3817219

b <- sum(asv_wtax_decon$value);b #3.799 million

## [1] 3788791

100*((b-a)/a)

## [1] -0.7447307

# Lost 0.47% of sequences from whole dataset.

## Subsample to clean ASVs
asv_wtax_wstats <- asv_wtax %>% 
  mutate(DECONTAM = case_when(
    FeatureID %in% list_of_contam_asvs ~ "FAIL",
    TRUE ~ "PASS"
  ))

Started with 17934 ASVs, post-decontamination, we have 17878 (a loss of 56 ASVs).

Data started with 3817219 sequences, after removing 56 ASVs, we have 3788791 total sequences. There was a total loss of 0.74% of sequences.

plot_grid(asv_wtax_wstats %>% 
  filter(value > 0) %>% 
  ggplot(aes(x = SAMPLE, fill = DECONTAM)) +
  geom_bar(stat = "count", width = 0.9, color = "black") +
  labs(y = "Total ASVs") +
  coord_flip() +
  theme_linedraw() +
  facet_grid(DATASET ~ ., scale = "free", space = "free") +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 1),
        strip.background = element_blank(), strip.text = element_text(color = "black"),
        legend.position = "bottom"),
  asv_wtax_wstats %>% 
  group_by(SAMPLE, SITE, DECONTAM, DATASET) %>% 
  summarise(SUM_SEQ_DOMAIN = sum(value)) %>% 
  ggplot(aes(x = SAMPLE, y = SUM_SEQ_DOMAIN, fill = DECONTAM)) +
  geom_bar(stat = "identity", color = "black", width = 0.9) +
  labs(y = "Total Sequences") +
  coord_flip() +
  theme_linedraw() +
  facet_grid(DATASET ~ ., scale = "free", space = "free") +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 1),
        strip.background = element_blank(), strip.text = element_text(color = "black"),
        legend.position = "bottom"),
  ncol = 2)

This plot shows the distribution of ASVs and sequences that failed or passed the decontamination step. Most obvious are the control samples that indicated the potentially contaminate ASVs.

Survey of sequences from in situ samples

# colnames(asv_wtax_wstats)
# unique(asv_wtax_wstats$SAMPLE)
sites <- c("Piccard", "VonDamm", "Axial", "GordaRidge")
asv_insitu <- asv_wtax_wstats %>% filter(Sample_or_Control != "Control") %>% 
       filter(SITE %in% sites) %>% 
       filter(!grepl("_expTf_", SAMPLE)) %>% 
       filter(value > 0) %>% 
       filter(DECONTAM == "PASS")

# Get quick stats on totals
sum(asv_insitu$value) # 3.8 million sequences

## [1] 3788791

length(unique(asv_insitu$FeatureID)) #12,378 ASVs

## [1] 12378

Final in situ dataset includes 3.79 million sequences and 12,378 ASVs total.

Additional sample QC, check replicates, and determine if replicates should be averaged.

plot_grid(asv_insitu %>% 
  group_by(SAMPLENAME, VENT, DATASET, Domain) %>% 
  summarise(seqsum_var = sum(value),
            asvcount_var = n()) %>% 
  pivot_longer(ends_with("_var"), names_to = "VARIABLE") %>% 
  ggplot(aes(x = SAMPLENAME, y = value, fill = Domain)) +
    geom_bar(color = "black", stat = "identity", position = "fill") +
    facet_grid(VARIABLE ~ DATASET, space = "free", scales = "free") +
  scale_y_continuous(expand = c(0,0)) +
  theme_linedraw() +
  scale_fill_brewer(palette = "Paired") +
  theme(strip.background = element_blank(), strip.text = element_text(color = "black"),
        axis.text.x = element_text(color = "black", angle = 90, hjust = 1, vjust = 0.5),
        legend.position = "bottom"),
  asv_insitu %>% 
  group_by(SAMPLENAME, VENT, DATASET, Domain) %>% 
  summarise(seqsum_var = sum(value),
            asvcount_var = n()) %>% 
  pivot_longer(ends_with("_var"), names_to = "VARIABLE") %>% 
  ggplot(aes(x = SAMPLENAME, y = value, fill = Domain)) +
    geom_bar(color = "black", stat = "identity", position = "stack") +
    facet_grid(VARIABLE ~ DATASET, space = "free_x", scales = "free") +
  scale_y_continuous(expand = c(0,0)) +
  theme_linedraw() +
  scale_fill_brewer(palette = "Paired") +
  theme(strip.background = element_blank(), strip.text = element_text(color = "black"),
        axis.text.x = element_text(color = "black", angle = 90, hjust = 1, vjust = 0.5),
        legend.position = "bottom"),
ncol = 2)

Base bar plot - taxonomy

asv_insitu %>% 
  filter(Domain == "Eukaryota") %>%
  # unite(SampleIdentifier, VENT, SAMPLETYPE, sep = " ", remove = FALSE) %>% 
  mutate(Supergroup = ifelse(is.na(Supergroup), "Unknown Eukaryota", Supergroup),
         Phylum = ifelse(is.na(Phylum), "Unknown", Phylum),
         Phylum = ifelse(Phylum == "Alveolata_X", "Ellobiopsidae", Phylum),
         Supergroup = ifelse(Supergroup == "Alveolata", paste(Supergroup, Phylum, sep = "-"), Supergroup)) %>% 
  group_by(FeatureID, Taxon, Domain, Supergroup, Phylum, Class, Order, Family, Genus, Species,
           VENT, SITE, SAMPLETYPE, YEAR, DATASET) %>% 
    summarise(SEQ_AVG_REP = mean(value)) %>% 
  ungroup() %>% 
  group_by(SITE, SAMPLETYPE, VENT, Supergroup) %>% 
    summarise(SEQ_SUM = sum(SEQ_AVG_REP)) %>% 
  ggplot(aes(x = VENT, y = SEQ_SUM, fill = Supergroup)) +
    geom_bar(stat = "identity", position = "stack", color = "black", width = 0.9) +
    facet_grid(. ~ SITE + SAMPLETYPE, scale = "free", space = "free") +
  theme_linedraw() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 1),
        strip.background = element_blank(), strip.text = element_text(color = "black")) +
  scale_y_continuous(expand = c(0,0)) +
  scale_fill_manual(values = c("#f1eef6", "#d7b5d8", "#df65b0", "#ce1256", "#fc9272", "#ef3b2c", 
    "#800026", "#fff7bc", "#fec44f", "#d95f0e", "#a63603", "#74c476", "#238b45", 
    "#00441b", "#7fcdbb", "#084081", "#c6dbef", "#2b8cbe", "#016c59", "#bcbddc", 
    "#807dba", "#54278f", "#bdbdbd", "black"))

Repeat taxonomy barplot, but with relative abundance

asv_insitu %>% 
  filter(Domain == "Eukaryota") %>%
  # unite(SampleIdentifier, VENT, SAMPLETYPE, sep = " ", remove = FALSE) %>% 
  mutate(Supergroup = ifelse(is.na(Supergroup), "Unknown Eukaryota", Supergroup),
         Phylum = ifelse(is.na(Phylum), "Unknown", Phylum),
         Phylum = ifelse(Phylum == "Alveolata_X", "Ellobiopsidae", Phylum),
         Supergroup = ifelse(Supergroup == "Alveolata", paste(Supergroup, Phylum, sep = "-"), Supergroup)) %>% 
  group_by(FeatureID, Taxon, Domain, Supergroup, Phylum, Class, Order, Family, Genus, Species,
           VENT, SITE, SAMPLETYPE, YEAR, DATASET) %>% 
    summarise(SEQ_AVG_REP = mean(value)) %>% 
  ungroup() %>% 
  group_by(SITE, SAMPLETYPE, VENT, Supergroup) %>% 
    summarise(SEQ_SUM = sum(SEQ_AVG_REP)) %>% 
  ggplot(aes(x = VENT, y = SEQ_SUM, fill = Supergroup)) +
    geom_bar(stat = "identity", position = "fill", color = "black", width = 0.9) +
    facet_grid(. ~ SITE + SAMPLETYPE, scale = "free", space = "free") +
  theme_linedraw() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 1),
        strip.background = element_blank(), strip.text = element_text(color = "black")) +
  scale_y_continuous(expand = c(0,0)) +
  scale_fill_manual(values = c("#f1eef6", "#d7b5d8", "#df65b0", "#ce1256", "#fc9272", "#ef3b2c", 
    "#800026", "#fff7bc", "#fec44f", "#d95f0e", "#a63603", "#74c476", "#238b45", 
    "#00441b", "#7fcdbb", "#084081", "#c6dbef", "#2b8cbe", "#016c59", "#bcbddc", 
    "#807dba", "#54278f", "#bdbdbd", "black"))

Removal of samples - QC

Filter samples so that the total number of sequences is greater than 20,000 sequences.

# head(asv_insitu)
# unique(asv_insitu$Sample_or_Control)
# hist(asv_insitu$value)
tmp <- (asv_insitu %>% 
          group_by(SAMPLE, SAMPLENAME) %>% 
          summarise(SUM = sum(value)) %>% 
        filter(SUM < 20000))
toofew <- as.character(unique(tmp$SAMPLE))
toofew

## [1] "Axial_Dependable_FS900_2013"         
## [2] "GordaRidge_BSW020_sterivex_2019_REPa"

Samples: Axial_Dependable_FS900_2013 and GordaRidge_BSW020_sterivex_2019_REPa removed due to too few sequences.

Sequence and ASV table summary

Final table reporting total sequences and ASVs for each sample.

asv_insitu_qc <- asv_insitu %>% 
  filter(!(SAMPLE %in% toofew)) %>% 
  filter(value > 0)

stats_seq_asv_postQC <- asv_insitu_qc %>% 
  group_by(SAMPLEID, VENT, DATASET, SITE, SAMPLETYPE, YEAR) %>%
    summarise(SEQ_SUM = sum(value),
     ASV_COUNT = n()) %>% 
  ungroup() %>% 
  gt(
    groupname_col = c("DATASET", "SITE", "YEAR"),
    rowname_col = "SAMPLEID"
  ) %>%
  tab_header(title = "Final sequence & ASV count")

stats_seq_asv_postQC

Final sequence & ASV count
	VENT	SAMPLETYPE	SEQ_SUM	ASV_COUNT
Axial - Axial - 2015
AnemonePlume	Anemone Plume	Plume	33599	474
BSW1500m	Deep seawater	Background	48604	492
FS915	Marker113	Vent	63629	655
GR - GordaRidge - 2019
BSW056	Deep seawater	Background	25095	498
BSW081	Shallow seawater	Background	38180	571
Plume001	Near vent BW	Background	104058	1203
Plume036	Candelabra Plume	Plume	57514	472
Plume096	Mt Edwards Plume	Plume	38211	582
Vent009	Mt Edwards	Vent	42591	245
Vent010	Mt Edwards	Vent	57564	558
Vent011	Mt Edwards	Vent	71398	664
Vent039	Venti Latte	Vent	45588	486
Vent040	Venti Latte	Vent	57186	688
Vent041	Venti Latte	Vent	64669	834
Vent086	Candelabra	Vent	57357	720
Vent087	Candelabra	Vent	45791	662
Vent088	Candelabra	Vent	40699	630
Vent105	Sir Ventsalot	Vent	57955	455
Vent106	Sir Ventsalot	Vent	41174	652
Vent107	Sir Ventsalot	Vent	43170	587
Axial - Axial - 2013
FS891	Anemone	Vent	38672	449
FS896	El Guapo	Vent	57160	804
FS898	N3Area	Vent	53609	608
FS902	Skadi	Vent	43916	40
FS903	Marker113	Vent	77470	684
FS904	Marker33	Vent	51169	473
FS905	Boca	Vent	190588	2173
Axial - Axial - 2014
FS906	Marker113	Vent	33394	333
FS908	Marker33	Vent	59502	771
FS910	Escargot	Vent	73927	988
MCR - VonDamm - 2020
NA	ArrowLoop	Vent	115164	927
NA	BSW	Background	118859	860
NA	MustardStand	Vent	153337	53
NA	OldManTree	Vent	144829	596
NA	Plume	Plume	150840	1163
NA	Rav2	Vent	286221	1785
NA	ShrimpHole	Vent	157739	958
NA	WhiteCastle	Vent	171000	32
NA	X18	Vent	166339	636
MCR - Piccard - 2020
NA	BSW	Background	138994	644
NA	LotsOShrimp	Vent	204055	686
NA	Plume	Plume	121820	332
NA	Shrimpocalypse	Vent	146098	705

# sum(asv_insitu_qc$value)
# length(unique(asv_insitu_qc$FeatureID))
gtsave(stats_seq_asv_postQC, filename = "../output-tables/seq_asv_count_postQC.html")

# tmp <- asv_insitu_qc %>% 
#   group_by(SAMPLEID, VENT, DATASET, SITE, SAMPLETYPE, YEAR) %>%
#     summarise(SEQ_SUM = sum(value),
#      ASV_COUNT = n())
# mean(tmp$SEQ_SUM); range(tmp$SEQ_SUM)
# mean(tmp$ASV_COUNT); range(tmp$ASV_COUNT)
# View(filter(tmp))

Assess ASV presence/absence

Set up analysis to classify each ASV based on distribution

# head(asv_insitu_qc)
# head(insitu_wide)
# unique(asv_insitu_qc$SAMPLETYPE)
# unique(asv_insitu_qc$SITE)

tax_asv_id <- asv_insitu_qc %>% 
  filter(value > 0) %>% #remove zero values
  select(FeatureID, SITE, SAMPLETYPE) %>% # isolate only ASVs that are PRESENT at a site and sampletype
  distinct() %>% #unique this, as presense = present in at least 1 rep (where applicable)
  unite(sample_id, SITE, SAMPLETYPE, sep = "_") %>% 
  # select(-SITE) %>% 
  # distinct() %>% 
  add_column(present = 1) %>%
  pivot_wider(names_from = sample_id, values_from = present, values_fill = 0) %>% 
  rowwise() %>% 
  mutate_at(vars(FeatureID), factor)

Is an ASV present only at the vent site? plume? or background? What about background and plume only?

library(purrr)
any_cols <- function(tax_asv_id) reduce(tax_asv_id, `|`)

asv_class <- tax_asv_id %>%
  mutate(vent = ifelse(any_cols(across(contains("_Vent"), ~ . > 0)), "VENT", ""),
         plume= ifelse(any_cols(across(contains("_Plume"), ~ . > 0)), "PLUME", ""),
         bsw = ifelse(any_cols(across(contains("_Background"), ~ . > 0)), "BSW", ""),
         ) %>% 
  unite(class_tmp, vent, plume, bsw, sep = "_", na.rm = TRUE) %>% 
  mutate(CLASS = case_when(
  class_tmp == "VENT__" ~ "Vent only",
  class_tmp == "_PLUME_" ~ "Plume only",
  class_tmp == "__BSW" ~ "Background only",
  class_tmp == "VENT__BSW" ~ "Vent & background",
  class_tmp == "VENT_PLUME_BSW" ~ "Vent, plume, & background",
  class_tmp == "VENT_PLUME_" ~ "Vent & plume",
  class_tmp == "_PLUME_BSW" ~ "Plume & background"
  )) %>% 
  select(FeatureID, CLASS) %>% distinct()

colnames(tax_asv_id)

##  [1] "FeatureID"             "GordaRidge_Vent"       "GordaRidge_Background"
##  [4] "Axial_Vent"            "VonDamm_Vent"          "GordaRidge_Plume"     
##  [7] "VonDamm_Plume"         "Piccard_Vent"          "Piccard_Background"   
## [10] "Axial_Plume"           "VonDamm_Background"    "Axial_Background"     
## [13] "Piccard_Plume"

Binary data frame with 1 indicating presence of ASV (rows) in a given sample (columns)

Depending on prevalence of ASV, assign groupings of location.

asv_class_SITE <- tax_asv_id %>%
  mutate(
    # mcr = ifelse(any_cols(across(contains("Piccard") | contains("VonDamm"), ~ . > 0)), "MCR", ""),
        picc = ifelse(any_cols(across(contains("Piccard"), ~ . > 0)), "Picc", ""),
        vd = ifelse(any_cols(across(contains("VonDamm"), ~ . > 0)), "VD", ""),
         axial = ifelse(any_cols(across(contains("Axial"), ~ . > 0)), "AxS", ""),
         gr = ifelse(any_cols(across(contains("Gorda"), ~ . > 0)), "GR", "")
         ) %>% 
  # unite(class_tmp, mcr, axial, gr, sep = "_", na.rm = TRUE) %>%
  unite(class_tmp, picc, vd, axial, gr, sep = "_", na.rm = TRUE) %>% 
  # unique(asv_class_SITE$class_tmp)
  mutate(SITE_CLASS = case_when(
  class_tmp == "___GR" ~ "Gorda Ridge only",
  class_tmp == "__AxS_" ~ "Axial only",
  class_tmp == "_VD__" ~ "Von Damm only",
  class_tmp == "Picc_VD__" ~ "Piccard & Von Damm",
  class_tmp == "Picc___" ~ "Piccard only",
  class_tmp == "Picc_VD_AxS_" ~ "MCR & Axial",
  class_tmp == "__AxS_GR" ~ "Axial & Gorda Ridge",
  class_tmp == "_VD__GR" ~ "Von Damm & Gorda Ridge",
  class_tmp == "_VD_AxS_GR" ~ "Von Damm, Axial, & Gorda Ridge",
  class_tmp == "_VD_AxS_" ~ "Von Damm & Axial",
  # class_tmp == "MCR__" ~ "Mid-Cayman Rise",
  class_tmp == "Picc_VD__GR" ~ "MCR & Gorda Ridge",
  class_tmp == "Picc__AxS_GR" ~ "Piccard, Axial, & Gorda Ridge",
  class_tmp == "Picc___GR" ~ "Piccard & Gorda Ridge",
  class_tmp == "Picc__AxS_" ~ "Piccard & Axial",
  class_tmp == "Picc_VD_AxS_GR" ~ "All sites"
  )) %>% 
  select(FeatureID, SITE_CLASS) %>% distinct()
# View(select(asv_class_SITE, SITE_CLASS) %>% distinct())

Combine together with original ASV table

insitu_asv_wClass <- asv_insitu_qc %>% 
  left_join(asv_class) %>% 
  left_join(asv_class_SITE)
# head(insitu_asv_wClass)

Visualize the total number of ASVs in background, plume, versus background.

# head(asv_insitu_qc)
# svg("bubbles.svg", h = 4, w = 8)
asv_insitu_qc %>% 
  select(DATASET, FeatureID, SAMPLETYPE) %>% 
  group_by(DATASET, SAMPLETYPE) %>% 
    summarise(COUNT = n_distinct(FeatureID)) %>% 
  ggplot(aes(x = DATASET, y = SAMPLETYPE, fill = SAMPLETYPE)) +
    geom_point(aes(size = COUNT), shape = 21, color = "black") +
    scale_size_continuous(range = c(4,20)) +
  scale_fill_viridis_d(option = "B") +
  theme_void() +
  theme(legend.position = "right",
        axis.text = element_text(color = "black"))

# dev.off()

Bubble plot reporting the total number of ASVs found in the vent, plume, versus background. At each site, the vent protist population had a higher total number of ASVs.

Repeat visualization by ASV distribution category.

# head(insitu_asv_wClass)
insitu_asv_wClass %>% 
  select(DATASET, FeatureID, SAMPLETYPE, CLASS) %>% 
  group_by(DATASET, SAMPLETYPE, CLASS) %>% 
    summarise(COUNT = n_distinct(FeatureID)) %>% 
  ggplot(aes(x = DATASET, y = SAMPLETYPE, fill = SAMPLETYPE)) +
    geom_point(aes(size = COUNT), shape = 21, color = "black") +
    scale_size_continuous(range = c(4,20)) +
  scale_fill_viridis_d(option = "B") +
  theme_void() +
  theme(legend.position = "right",
        axis.text.x = element_text(color = "black"),
        axis.title.y = element_blank()) +
  facet_grid(SAMPLETYPE + CLASS ~ ., scales = "free", space = "free") +
  labs(x = "", y = "", title = "Total number of ASVs by distribution & sample type")

Repeated bubble plot reports the total number of ASVs in the vent, plume, and background - but now further separated by distribution (i.e., if an ASV was found only in the vent and plume = “Vent & plume”). The largest portion of ASVs were found only at the vent sites (Vent only).

Categories for ASV distribution:

unique(insitu_asv_wClass$CLASS)

## [1] "Vent only"                 "Background only"          
## [3] "Vent & background"         "Vent, plume, & background"
## [5] "Plume only"                "Vent & plume"             
## [7] "Plume & background"

unique(insitu_asv_wClass$SITE_CLASS)

##  [1] "Gorda Ridge only"               "Axial only"                    
##  [3] "Von Damm only"                  "Piccard & Von Damm"            
##  [5] "Piccard only"                   "MCR & Axial"                   
##  [7] "Axial & Gorda Ridge"            "Von Damm & Gorda Ridge"        
##  [9] "Von Damm, Axial, & Gorda Ridge" "Von Damm & Axial"              
## [11] "All sites"                      "MCR & Gorda Ridge"             
## [13] "Piccard, Axial, & Gorda Ridge"  "Piccard & Gorda Ridge"         
## [15] "Piccard & Axial"

length(unique(insitu_asv_wClass$FeatureID))

## [1] 12375

sum(insitu_asv_wClass$value)

## [1] 3788734

Save working ASV table

Checkpoint to save working dataframes.

save(asv_insitu, asv_insitu_qc, insitu_asv_wClass, file = "../data-input/asv-tables-processed-27012022.RData")

Process QIMME2 output, run decontam

Sarah Hu

1/26/2022